Methods	RCT with randomisation of individual women. Sequentially numbered opaque envelopes, using pseudo-random number generator
Participants	Inclusion criteria • Women with signs of incomplete miscarriage. • Diagnosis confirmed by ultrasound. • 1st trimester; good general health; no allergy to misoprostol; good access to emergency facilities. • N = 169 women. Exclusion criteria • None specified.
Interventions	Intervention: oral misoprostol - 600 ug, single dose (N = 86). Comparison: oral misoprostol - 1200 ug, 2 doses, 4 hours apart (N = 83)
Outcomes	Complete miscarriage at 48 hours; surgical evacuation; side effects and acceptability • Assessed by ultrasound at 48 hours. • If miscarriage not complete at 48 hours, women were given the option to wait additional 5 days (1 week from misoprostol administration) to see if miscarriage would be complete without further intervention. If miscarriage not complete after 1 week or if woman refused extension, then she underwent surgical evacuation according to standard practice.
Notes	1. Setting: 2 teaching hospitals in Bangkok, Thailand. 2. Additional outcomes assessed but not pre-specified in the review: bleeding (heavy, normal, spotting); pain; fever; medically necessary interventions; satisfied or very satisfied with treatment.
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“Pseudo-random number generator in SSPS 9.0.”
Allocation concealment (selection bias)	Low risk	Women given the next “…sequentially numbered opaque envelope; the number in the envelope became her study identification number”
Blinding (performance bias and detection bias) All outcomes	High risk	“Neither the provider nor the woman was blinded to the treatment regimes.”
Incomplete outcome data (attrition bias) All outcomes	Low risk	2 women in single-dose group and 1 woman in double-dose group were lost to follow up. 1.8% of total, so no real impact.
Selective reporting (reporting bias)	Unclear risk	Appears to be free of selective reporting bias but we did not assess the trial protocol
Other bias	Low risk	Appears to be free of other reporting bias.